1. Field of the Invention
The present invention relates to a method of semiconductor manufacturing in which P-type doping is accomplished by the implantation of ion beams formed from ionized boron hydride molecules, said ions being of the form BnHx+ and BnHx− where 10≦n≦100 and 0≦x≦n+4.
2. Description of the Prior Art
The Ion Implantation Process
The fabrication of semiconductor devices involves, in part, the introduction of impurities into the semiconductor substrate to form doped regions. The impurity elements are selected to bond appropriately with the semiconductor material so as to create electrical carriers, thus altering the electrical conductivity of the semiconductor material. The electrical carriers can either be electrons (generated by N-type dopants) or holes (generated by P-type dopants). The concentration of dopant impurities so introduced determines the electrical conductivity of the resultant region. Many such N- and P-type impurity regions must be created to form transistor structures, isolation structures and other such electronic structures which function collectively as a semiconductor device.
The conventional method of introducing dopants into a semiconductor substrate is by ion implantation. In ion implantation, a feed material containing the desired element is introduced into an ion source and energy is introduced to ionize the feed material, creating ions which contain the dopant element (for example, in silicon the elements 75As, 31P, and 121Sb are donors or N-type dopants, while 11B and 115In are acceptors or P-type dopants). An accelerating electric field is provided to extract and accelerate the typically positively-charged ions, thus creating an ion beam (in certain cases, negatively-charged ions may be used instead). Then, mass analysis is used to select the species to be implanted, as is known in the art, and the mass-analyzed ion beam may subsequently pass through ion optics which alter its final velocity or change its spatial distribution prior to being directed into a semiconductor substrate or workpiece. The accelerated ions possess a well-defined kinetic energy which allows the ions to penetrate the target to a well-defined, predetermined depth at each energy value. Both the energy and mass of the ions determine their depth of penetration into the target, with higher energy and/or lower mass ions allowing deeper penetration into the target due to their greater velocity. The ion implantation system is constructed to carefully control the critical variables in the implantation process, such as the ion energy, ion mass, ion beam current (electrical charge per unit time), and ion dose at the target (total number of ions per unit area that penetrate into the target). Further, beam angular divergence (the variation in the angles at which the ions strike the substrate) and beam spatial uniformity and extent must also be controlled in order to preserve semiconductor device yields.
A key process of semiconductor manufacturing is the creation of P-N junctions within the semiconductor substrate. This requires the formation of adjacent regions of P-type and N-type doping. An important example of the formation of such a junction is the implantation of P-type dopant into a semiconductor region already containing a uniform distribution of N-type dopant. In this case, an important parameter is the junction depth, which is defined as the depth from the semiconductor surface at which the P-type and N-type dopants have equal concentrations. This junction depth is a function of the implanted dopant mass, energy and dose.
An important aspect of modern semiconductor technology is the continuous evolution to smaller and faster devices. This process is called scaling. Scaling is driven by continuous advances in lithographic process methods, allowing the definition of smaller and smaller features in the semiconductor substrate which contains the integrated circuits. A generally accepted scaling theory has been developed to guide chip manufacturers in the appropriate resize of all aspects of the semiconductor device design at the same time, i.e., at each technology or scaling node. The greatest impact of scaling on ion implantation process is the scaling of junction depths, which requires increasingly shallow junctions as the device dimensions are decreased. This requirement for increasingly shallow junctions as integrated circuit technology scales translates into the following requirement: ion implantation energies must be reduced with each scaling step. The extremely shallow junctions called for by modern, sub-0.13 micron devices are termed “Ultra-Shallow Junctions”, or USJ.
Physical Limitations on Low-Enemy Beam Transport
Due to the aggressive scaling of junction depths in CMOS processing, the ion energy required for many critical implants has decreased to the point that conventional ion implantation systems, originally developed to generate much higher energy beams, deliver much reduced ion currents to the wafer, reducing wafer throughput. The limitations of conventional ion implantation systems at low beam energy are most evident in the extraction of ions from the ion source, and their subsequent transport through the implanter's beam line. Ion extraction is governed by the Child-Langmuir relation, which states that the extracted beam current density is proportional to the extraction voltage (i.e., beam energy at extraction) raised to the 3/2 power. FIG. 2 is a graph of maximum extracted boron beam current versus extraction voltage. For simplicity, an assumption has been made that only 11B+ ions are present in the extracted beam. FIG. 2 shows that as the energy is reduced, extraction current drops quickly. In a conventional ion implanter, this regime of “extraction-limited” operation is seen at energies less than about 10 keV. Similar constraints affect the transport of the low-energy beam after extraction. A lower energy ion beam travels with a smaller velocity, hence for a given value of beam current the ions are closer together, i.e., the ion density increases. This can be seen from the relation J=f ηev, where J is the ion beam current density in mA/cm2, η is the ion density in ions/cm−3, e is the electronic charge (=6.02×10−19 Coulombs), and v is the average ion velocity in cm/s. In addition, since the electrostatic forces between ions are inversely proportional to the square of the distance between them, electrostatic repulsion is much stronger at low energy, resulting in increased dispersion of the ion beam. This phenomenon is called “beam blow-up”, and is the principal cause of beam loss in low-energy transport. While low-energy electrons present in the implanter beam line tend to be trapped by the positively-charged ion beam, compensating for space-charge blow-up during transport, blow-up nevertheless still occurs, and is most pronounced in the presence of electrostatic focusing lenses, which tend to strip the loosely-bound, highly mobile compensating electrons from the beam. In particular, severe extraction and transport difficulties exist for light ions, such as the P-type dopant boron, whose mass is only 11 amu. Being light, boron atoms penetrate further into the substrate than other atoms, hence the required implantation energies for boron are lower than for the other implant species. In fact, extremely low implantation energies of less than 1 keV are being required for certain leading edge USJ processes. In reality, most of the ions extracted and transported from a typical BF3 source plasma are not the desired ion 11B+, but rather ion fragments such as 19F+ and 49BF2+; these serve to increase the charge density and average mass of the extracted ion beam, further increasing space-charge blow-up. For a given beam energy, increased mass results in a greater beam perveance; since heavier ions move more slowly, the ion density η increases for a given beam current, increasing space charge effects in accordance with the discussion above.
Molecular Ion Implantation
One way to overcome the limitations imposed by the Child-Langmuir relation discussed above is to increase the transport energy of the dopant ion by ionizing a molecule containing the dopant of interest, rather than a single dopant atom. In this way, while the kinetic energy of the molecule is higher during transport, upon entering the substrate, the molecule breaks up into its constituent atoms, sharing the energy of the molecule among the individual atoms according to their distribution in mass, so that the dopant atom's implantation energy is much lower than the original transport kinetic energy of the molecular ion. Consider the dopant atom “X” bound to a radical “Y” (disregarding for purposes of discussion the issue of whether “Y” affects the device-forming process). If the ion XY+ were implanted in lieu of X+, then XY+ must be extracted and transported at a higher energy, increased by a factor equal to the mass of XY divided by the mass of X; this ensures that the velocity of X in either case is the same. Since the space-charge effects described by the Child-Langmuir relation discussed above are super-linear with respect to ion energy, the maximum transportable ion current is increased. Historically, the use of polyatomic molecules to ameliorate the problems of low energy implantation is well known in the art. A common example has been the use of the BF2+ molecular ion for the implantation of low-energy boron, in lieu of B+. This process dissociates BF3 feed gas to the BF2+ ion for implantation. In this way, the ion mass is increased to 49 AMU, allowing an increase of the extraction and transport energy by more than a factor of 4 (i.e., 49/11) over using single boron atoms. Upon implantation, however, the boron energy is reduced by the same factor of (49/11). It is worthy of note that this approach does not reduce the current density in the beam, since there is only one boron atom per unit charge in the beam. In addition, this process also implants fluorine atoms into the semiconductor substrate along with the boron, an undesirable feature of this technique since fluorine has been known to exhibit adverse effects on the semiconductor device.
Cluster Implantation
In principle, a more effective way to increase dose rate than by the XY+ model discussed above is to implant clusters of dopant atoms, that is, molecular ions of the form XnYm+, where n and m are integers and n is greater than one. Recently, there has been seminal work using decaborane as a feed material for ion implantation. The implanted particle was a positive ion of the decaborane molecule, B10H14, which contains 10 boron atoms, and is therefore a “cluster” of boron atoms. This technique not only increases the mass of the ion and hence the transport ion energy, but for a given ion current, it substantially increases the implanted dose rate, since the decaborane ion B10Hx+ has ten boron atoms. Importantly, by significantly reducing the electrical current carried in the ion beam (by a factor of 10 in the case of decaborane ions) not only are beam space-charge effects reduced, increasing beam transmission, but wafer charging effects are reduced as well. Since positive ion bombardment is known to reduce device yields by charging the wafer, particularly damaging sensitive gate isolation, such a reduction in electrical current through the use of cluster ion beams is very attractive for USJ device manufacturing, which must increasingly accommodate thinner gate oxides and exceedingly low gate threshold voltages. Thus, there is a critical need to solve two distinct problems facing the semiconductor manufacturing industry today: wafer charging, and low productivity in low-energy ion implantation. As we will show later in this document, the present invention proposes to further increase the benefits of cluster implantation by using significantly larger boron hydride clusters having n>10. In particular, we have implanted the B18Hx+ ion, and further propose to implant the B36Hx+ ion, using the solid feed material octadecaborane, or B18H22. We will present first results showing that this technology is a significant advance over previous efforts in boron cluster implantation.
Ion Implantation Systems
Ion implanters have historically been segmented into three basic categories: high current, medium current, and high energy implanters. Cluster beams are useful for high current and medium current implantation processes. In particular, today's high current implanters are primarily used to form the low energy, high dose regions of the transistor such as drain structures and doping of the polysilicon gates. They are typically batch implanters, i.e., processing many wafers mounted on a spinning disk, the ion beam remaining stationary. High current transport systems tend to be simpler than medium current transport systems, and incorporate a large acceptance of the ion beam. At low energies and high currents, prior art implanters produce a beam at the substrate which tends to be large, with a large angular divergence (e.g., a half-angle of up to seven degrees). In contrast, medium current implanters typically incorporate a serial (one wafer at a time) process chamber, which offers a high tilt capability (e.g., up to 60 degrees from the substrate normal). The ion beam is typically electromagnetically or electrodynamically scanned across the wafer at a high frequency, up to about 2 kiloHertz in one dimension (e.g., laterally) and mechanically scanned at a low frequency of less than 1 Hertz in an orthogonal direction (e.g., vertically), to obtain areal coverage and provide dose uniformity over the substrate. Process requirements for medium current implants are more complex than those for high current implants. In order to meet typical commercial implant dose uniformity and repeatability requirements of a variance of only a few percent, the ion beam must possess excellent angular and spatial uniformity (angular uniformity of beam on wafer of ≦1 deg, for example). Because of these requirements, medium current beam lines are engineered to give superior beam control at the expense of reduced acceptance. That is, the transmission efficiency of the ions through the implanter is limited by the emittance of the ion beam. Presently, the generation of higher current (about 1 mA) ion beams at low (<10 keV) energy is problematic in serial implanters, such that wafer throughput is unacceptably low for certain lower energy implants (for example, in the creation of source and drain structures in leading edge CMOS processes). Similar transport problems also exist for batch implanters (processing many wafers mounted on a spinning disk) at the low beam energies of <5 keV per ion.
While it is possible to design beam transport optics which are nearly aberration-free, the ion beam characteristics (spatial extent, spatial uniformity, angular divergence and angular uniformity) are nonetheless largely determined by the emittance properties of the ion source itself (i.e., the beam properties at ion extraction which determine the extent to which the implanter optics can focus and control the beam as emitted from the ion source). The use of cluster beams instead of monomer beams can significantly enhance the emittance of an ion beam by raising the beam transport energy and reducing the electrical current carried by the beam. However, prior art ion sources for ion implantation are not effective at producing or preserving ionized clusters of the required N- and P-type dopants. Thus, there is a need for cluster ion and cluster ion source technology in order to provide a better-focused, more collimated and more tightly controlled ion beam on target, and in addition to provide higher effective dose rates and higher throughputs in semiconductor manufacturing.
An alternative approach to beam line ion implantation for the doping of semiconductors is so-called “plasma immersion”. This technique is known by several other names in the semiconductor industry, such as PLAD (PLAsma Doping), PPLAD (Pulsed PLAsma Doping, and PI3 (Plasma Immersion Ion Implantation). Doping using these techniques requires striking a plasma in a large vacuum vessel that has been evacuated and then backfilled with a gas containing the dopant of choice such as boron triflouride, diborane, arsine, or phosphine. The plasma by definition has positive ions, negative ions and electrons in it. The target is then biased negatively thus causing the positive ions in the plasma to be accelerated toward the target. The energy of the ions is described by the equation U=QV, where U is the kinetic energy of the ions, Q is the charge on the ion, and V is the bias on the wafer. With this technique there is no mass analysis. All positive ions in the plasma are accelerated and implanted into the wafer. Therefore extremely clean plasma must be generated. With this technique of doping a plasma of diborane, phosphine or arsine gas is formed, followed by the application of a negative bias on the wafer. The bias can be constant in time, time-varying, or pulsed. Dose can be parametrically controlled by knowing the relationship between pressure of the vapor in the vessel, the temperature, the magnitude of the biasing and the duty cycle of the bias voltage and the ion arrival rate on the target. It is also possible to directly measure the current on the target. While Plasma Doping is considered a new technology in development, it is attractive since it has the potential to reduce the per wafer cost of performing low energy, high dose implants, particularly for large format (e.g., 300 mm) wafers. In general, the wafer throughput of such a system is limited by wafer handling time, which includes evacuating the process chamber and purging and re-introducing the process gas each time a wafer or wafer batch is loader into the process chamber. This requirement has reduced the throughput of Plasma Doping systems to about 100 wafers per hour (WPH), well below the maximum mechanical handling capability of beamline ion implantation systems, which can process over 200 WPH.
Negative Ion Implantation
It has recently been recognized (see, for example, Junzo Ishikawa et al., “Negative-Ion Implantation Technique”, Nuclear Instruments and Methods in Physics Research B 96 (1995) 7-12.) that implanting negative ions offers advantages over implanting positive ions. One very important advantage of negative ion implantation is to reduce the ion implantation-induced surface charging of VLSI devices in CMOS manufacturing. In general, the implantation of high currents (on the order of 1 mA or greater) of positive ions creates a positive potential on the gate oxides and other components of the semiconductor device which can easily exceed gate oxide damage thresholds. When a positive ion impacts the surface of a semiconductor device, it not only deposits a net positive charge, but liberates secondary electrons at the same time, multiplying the charging effect. Thus, equipment vendors of ion implantation systems have developed sophisticated charge control devices, so-called electron flood guns, to introduce low-energy electrons into the positively-charged ion beam and onto the surface of the device wafers during the implantation process. Such electron flood systems introduce additional variables into the manufacturing process, and cannot completely eliminate yield losses due to surface charging. As semiconductor devices become smaller and smaller, transistor operating voltages and gate oxide thicknesses become smaller as well, reducing the damage thresholds in semiconductor device manufacturing, further reducing yield. Hence, negative ion implantation potentially offers a substantial improvement in yield over conventional positive ion implantation for many leading-edge processes. Unfortunately, this technology is not yet commercially available, and indeed negative ion implantation has not to the author's knowledge been used to fabricate integrated circuits, even in research and development.